Вы находитесь на странице: 1из 2

DOI: 10.1111/jbg.

12361

EDITORIAL

The fundamental theorem of natural selection


This is the last editorial celebrating Fisher’s classic and belongs to a class with 1927 value vi, then Fisher’s implicit
hugely influential 1918 paper (Trans. R. Soc. Edinburgh 52, definition of its fitness mi at that time is
399–433) in its centenary year. Selection was mentioned in
Wi − vi
1918 as a possible confounder of measured correlations, but mi = ,
vi
the main aim was to reconcile biometrics with Mendelism. In
The Genetical Theory of Natural Selection in 1930, however,
Fisher wanted to render mathematically respectable the idea which is therefore the proportional advantage an individual
that natural selection improved organisms and created biolog- has over the average its age class would have in a static popula-
ical design. Fisher had taken with him as an undergraduate to tion. Fisher’s fitness has a number of remarkable properties: (a)
Cambridge a set of Darwin’s complete works, and through- the mean fitness within each age class is the same; (b) despite
out his life wrote eloquently and influentially about the broad being a weighted sum of reproductive value (“downstairs”), the
biological (and more controversially social) implications of mean fitness equals the Malthusian parameter, that is the growth
Darwinian thinking. rate of the population as measured by the Euler–Lotka equation
When natural selection became the focus of Fisher’s think- in its exponent (“upstairs”); and (c) in a population in demo-
ing, some of the simplifying assumptions of 1918 became too graphic equilibrium, the mean fitness is therefore zero. The up-
simple. In particular, Fisher (1927, Eug. Rev. 19, 103–108) stairs/downstairs duality lies at the heart of the theory and binds
confronted the complications of age structure in a study of together evolutionary and ecological concepts of fitness.
human life tables and developed the concept of reproductive A second question is: How can mean fitness always be going
value. Each age class was assigned its own numerical value, up? Reconciling the non‐negative value from the fundamental
which represented relative contributions of typical individ- theorem with an ecological necessity of roughly constant fit-
uals of the different classes to the asymptotic gene pool. He ness turns out to be relatively simple. Fisher was usually clear
went on to develop, for age‐structured populations, the funda- that only part of the rate of change equals what we would now
mental theorem of natural selection as the jewel in the crown call the additive genetic variance in fitness, namely that part
of his 1930 book, namely “The rate of increase in fitness of due to natural selection, and within a dozen pages of the theo-
any organism at any time is equal to its genetic variance in rem showed he appreciated this second puzzle, by presenting a
fitness at that time” (Fisher, 1930, page 35). model of other components that can, and usually will, leave the
But, what is fitness? Recent work (Grafen, 2015, Am. Nat. total rate of change at about zero. Fisher preferred brevity over
186, 1–14) shows that Fisher’s concept of fitness is not just [i.e. a space to make two words "over" and then "complete-
“a parameter in a model” (pace Ewens, 2004 Mathematical ness"] completeness in the famous verbal statement, which
Population Genetics I. Theoretical Introduction. Springer). may be responsible for the theorem’s evident power to inspire,
Each individual has a fitness at each moment in time, not but proved to be catastrophic for readers’ understanding of it.
a single “tombstone” figure. It is convenient to discuss dis- Biologists in traditions inspired by the theorem mainly
crete time, as the principles are the same. Using the 1927 cited Hamilton (1964, J. Theor. Biol 7, 1–52) and Williams
per capita class‐based reproductive values, the reproductive (1966, ibid.), who themselves cited Fisher. The shining ex-
value the individual produces this period comes under two ample of mathematical support for this work is the oeuvre of
headings: half of the reproductive value of its expected num- Charlesworth from 1970 (Theor. Pop. Biol. 1, 352–370) on-
ber of offspring (age zero), and the individual’s own chance wards (now summarized in Charlesworth, 1994 Evolution in
of survival times her reproductive value (at 1 year older, Age‐Structured Populations, 2nd ed., Cambridge University
of course). This is like the usual Leslie matrix calculation, Press), though its place in the standard textbook (Ewens,
except that here it is performed on an individual basis, al- 2004, ibid.) reveals the widespread view among mathemat-
lowing an individual’s genotype to influence offspring num- ical population geneticists. They had held for many decades
ber and survival. Williams (1966, Am. Nat, 100, 687–690) an extremely dim view of the fundamental theorem itself
founded life‐history theory with this concept, which we will (e.g. Charlesworth, 1970, ibid.; Karlin and Feldman, 1970,
call “Williams’ reproductive value” to distinguish it from Theor. Pop. Biol. 7, 364–398; Karlin, 1975, Theor. Pop. Biol.
the 1927 class‐based values. Indeed, if individual i at some 1, 39–71), though some showed very little sign of having
particular moment has Williams’ reproductive value Wi and read the original. Geneticists began to come to terms with the

J Anim Breed Genet. 2018;135:393–394. wileyonlinelibrary.com/journal/jbg © 2018 Blackwell Verlag GmbH     393 |
|
394       EDITORIAL

theorem in the 1980s, and it is still controversial—see Ewens very basic and important idea with a fruitful way of look-
and Lessard (2015, Theor. Pop. Biol. 104, 59–67) and Grafen ing at the subject, and be capable of absorbing further intel-
(2018, J. Theor. Biol. 456, 175–189) for the latest news and lectual developments. Fisher had called in his 1930 preface
for references to the history. for a body of mathematical work in biology to match that
The eclipse of the theorem mattered because biology was in physics, presumably mindful that physics is a very broad
a very broad subject, with theory and applications, and with subject that is bound together by standard theoretical nota-
molecules, cells, organisms and communities. Holding it tions and results that are part of the professional training of
all together requires general principles, and the fundamen- all physicists. Biology still lacks any similar body of didac-
tal theorem touched more parts of biology than any other. tic material: the fundamental theorem is in my view still the
Fisher himself (1930, ibid, p37) likened the theorem to the place to start.
Second Law of Thermodynamics and wrote “It is not a little I end with two reflections. The centrality of applied work
instructive that so similar a law should hold the supreme po- in the study of natural selection began with its invention by
sition among the biological sciences.” This view has not been Darwin, for whom the breeding of pigeons and cattle was a
widely shared, but consider the remarks in an acceptance lec- central inspiration. It continued with Fisher, whose invention
ture of a Nobel Prize in chemistry, which begin by referring of reproductive value in 1927 came from work on an empiri-
to an inequality involving the change in entropy over time: cal life table for humans. And Robertson (1966, Anim. Prod.
8, 95–108) introduced his “secondary theorem of natural se-
The inequality in eq. 3, which does not involve lection” while writing about the culling process in cattle. The
the total differentiation of a function, does not high‐flying theoretical imagination benefits from concentra-
in general permit one to define a Lyapunov tion on cold hard facts.
function. Before we come back to this ques- Second, the fundamental theorem exhibits the scientific
tion, I should emphasize that 150 years after its contribution of Fisher and of his 1918 paper. The very term
formulation the second law of thermodynam- “variance” was introduced in the 1918 paper, and Fisher could
ics still appears to be more a program than a understandably have felt he owned natural selection in a very
well‐defined theory in the usual sense, as noth- special way when a variance component turned up in the fun-
ing precise (except the sign) is said about the damental theorem. We should be aiming to catch up on lost
S production. Even the range of validity of this time by embracing Fisher’s result with the same enthusiasm!
inequality is left unspecified. (Prigogine, 1978, Alan Grafen
Science 201, 777–785, page 778) Zoology Department and St John’s College, Oxford, UK

The significant point here is the realistic appreciation of Email: alan.grafen@sjc.ox.ac.uk


the true role of a foundational principle. It must capture a

Вам также может понравиться